Field of the Invention
The present invention relates to an image capturing apparatus, a control method thereof and a storage medium.
Description of the Related Art
Generally, in an image capturing apparatus provided with an image sensor such as a CCD or CMOS image sensor, correction processing is performed for phenomena influencing image quality, such as fixed-pattern noise, shading, and flickers that occur inside the image sensor. In correction processing, correction values for the respective phenomena are generated from image signals output from the image sensor, and the image signals are corrected using the generated correction values.
Incidentally, in recent image sensors, demand for an increased number of pixels and high speed image capturing has further increased, and there has been demand to perform signal processing at a high speed. Therefore, it has become necessary to increase the speed of a circuit for processing analog signals output from pixels. Accordingly, various parallel reading processing techniques have been proposed, such as simultaneously reading out analog signals output from pixels for each column, and in subsequent horizontal transfer, providing a plurality of amplifiers as output units, instead of intensively using one amplifier. Such parallel reading processing techniques have made it possible to increase the number of pixels and to capture images at a high speed. However, even if respective circuits for performing parallel reading processing are designed similarly in terms of layout design, variation in the properties thereof arise due to variation in manufacturing processes and the like. These property variations appear as stripe-shaped fixed-pattern noise in output images.
Accordingly, for example, Japanese Patent No. 4144517 discloses a method for calculating correction values for correcting fixed-pattern noise using a plurality of frames, with fixed-pattern noise detection rows for detecting fixed-pattern noise provided on an image sensor. In this method, even in the case where a sufficient number of fixed-pattern noise detection rows cannot be obtained in order to increase the speed of reading out from the image sensor, it is possible to improve the accuracy of correction patterns for correcting fixed-pattern noise by increasing the number of frames.
Furthermore, in Japanese Patent No. 5132386, a method is proposed in which detection of fixed-pattern noise is performed in advance for each of a plurality of reading methods of an image sensor and correction values are then generated so that the correction values are switched at the time of switching the reading methods.
However, in the method described in Japanese Patent No. 4144517, correction values are calculated from the beginning in the case where fixed-pattern shapes change. Examples of methods for reading out pixel signals of an image sensor include a combined reading method in which the signals of multiple pixels in a horizontal direction and in a vertical direction are combined and then read out, and a non-combined reading method in which the signals of all pixels are read out one pixel at a time. The combined reading method is used for generating a moving image for which fast readout is required, and the non-combined reading method is used for generating a high definition still image. Here, in the case where horizontal-direction combining is performed in the combined reading method, fixed-pattern noise is totaled to appear as one column, resulting in fixed-pattern shapes different from those caused in the case of the non-combined reading method.
When the reading method is switched, the number of samples of the fixed-pattern detection rows decreases and the S/N ratio of correction values becomes smaller. Therefore, there has been a problem in that fixed-pattern noise in images immediately after switching the reading methods could not be corrected accurately.
Also, in a method described in Japanese Patent No. 5132386, although switching of reading methods is handled, detection is performed in advance to generate correction values, and therefore there is a problem in that correction cannot be performed accurately in the case where fixed-pattern noise shapes change due to temperature change or the like.